OpenAI Agents SDK

Integrate Agent Diff with OpenAI Agents SDK

Overview

The OpenAI Agents SDK is a lightweight framework for building AI agents. Agent Diff provides native integration through code executor tools.

Installation

pip install agent-diff openai-agents

Basic Integration

from agent_diff import AgentDiff, PythonExecutorProxy, create_openai_tool
from agents import Agent, Runner

# Initialize client and environment
client = AgentDiff()
env = client.init_env(
    templateService="slack",
    templateName="slack_default",
    impersonateUserId="U01AGENBOT9"
)

# Create executor and tool
python_executor = PythonExecutorProxy(env.environmentId, base_url=client.base_url)
python_tool = create_openai_tool(python_executor)

# Create agent with the tool
agent = Agent(
    name="Slack Assistant",
    instructions="""Use execute_python tool to interact with Slack API at 
    https://slack.com/api/*. Authentication is handled automatically via proxy.
    Leave a placeholder credential where you would add a real token.""",
    tools=[python_tool]
)

# Start run and execute
run = client.start_run(envId=env.environmentId)
response = await Runner.run(agent, "Post 'Hello World!' to #general")

# Get diff
diff = client.diff_run(runId=run.runId)
print(diff.diff['inserts'])

# Cleanup
client.delete_env(envId=env.environmentId)

Using Bash Executor

For agents that prefer curl commands:

from agent_diff import BashExecutorProxy, create_openai_tool

bash_executor = BashExecutorProxy(env.environmentId, base_url=client.base_url)
bash_tool = create_openai_tool(bash_executor)

agent = Agent(
    name="Slack Assistant",
    instructions="""Use execute_bash tool with curl to interact with Slack API 
    at https://slack.com/api/*. Authentication is handled automatically.""",
    tools=[bash_tool]
)

Running Evaluations

from agent_diff import AgentDiff, PythonExecutorProxy, create_openai_tool
from agents import Agent, Runner

client = AgentDiff()

# Get test suite
suites = client.list_test_suites(name="Slack Bench")
suite = client.get_test_suite(suites.testSuites[0].id, expand=True)

results = []

for test in suite.tests:
    # Create environment for this test
    env = client.init_env(testId=test.id)
    
    # Start run
    run = client.start_run(envId=env.environmentId, testId=test.id)
    
    # Create executor and agent
    executor = PythonExecutorProxy(env.environmentId, client.base_url)
    agent = Agent(
        name="Slack Assistant",
        tools=[create_openai_tool(executor)],
        instructions="Use execute_python to interact with Slack at https://slack.com/api/*"
    )
    
    # Run agent with test prompt
    await Runner.run(agent, test.prompt)
    
    # Evaluate
    result = client.evaluate_run(runId=run.runId)
    
    results.append({
        "test": test.name,
        "passed": result.passed,
        "score": result.score
    })
    
    # Cleanup
    client.delete_env(envId=env.environmentId)

# Print results
passed = sum(1 for r in results if r['passed'])
print(f"\n Results: {passed}/{len(results)} tests passed")
for r in results:
    status = "✓" if r['passed'] else "✗"
    print(f"  {status} {r['test']}")

Multiple Tools

You can combine Python and Bash executors:

python_executor = PythonExecutorProxy(env.environmentId, client.base_url)
bash_executor = BashExecutorProxy(env.environmentId, client.base_url)

agent = Agent(
    name="Multi-Tool Agent",
    instructions="""You have two tools available:
    - execute_python: For Python scripts using requests library
    - execute_bash: For curl commands
    
    Use whichever is more appropriate for the task.
    Both route to Slack API at https://slack.com/api/*""",
    tools=[
        create_openai_tool(python_executor),
        create_openai_tool(bash_executor)
    ]
)

Linear API Example

env = client.init_env(
    templateService="linear",
    templateName="linear_expanded",
    impersonateEmail="[e-mail]"
)

python_executor = PythonExecutorProxy(env.environmentId, client.base_url)

agent = Agent(
    name="Linear Assistant",
    instructions="""Use execute_python to interact with Linear GraphQL API at 
    https://api.linear.app/graphql. Authentication is automatic.""",
    tools=[create_openai_tool(python_executor)]
)

run = client.start_run(envId=env.environmentId)
await Runner.run(agent, "Create a new issue titled 'Fix login bug' in the Engineering team")

diff = client.diff_run(runId=run.runId)
print(diff.diff['inserts'])

Next Steps

LangChain

Use Agent Diff with LangChain

smolagents

Use with HuggingFace smolagents